A method for improving the estimation of an existence probability of objects. The objects are detected using sensors installed in a vehicle and/or an infrastructure component. Each tracker of a sensor and/or a sensor group estimates a status of an object and its existence probability using a detection probability model. The detected objects are merged in a fusion list, and each object is assigned a state and an existence probability. Each object of the fusion list is assigned existence probabilities. Each object of the fusion list is assigned additional information indicating which sensor and/or which sensor group has/have detected the respective object in the last measuring cycle. At least sensor-specific and/or sensor-group-specific existence probabilities of fused existence probabilities and the sensor detection probability are compared in a crosscheck, and false negative cases and false positive cases are ascertained for each sensor and/or sensor groups.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
2. The method as recited in claim 1, wherein false negative cases or false positive cases assigned to each object are stored, in a manner specific to the sensor or the sensor group, in a false negative list or a false positive list as a function of the position of the object in the fusion list.
This method improves the estimation of object existence probabilities. It involves detecting objects with vehicle/infrastructure sensors, estimating their existence probabilities using detection models, merging them into a fusion list, and tracking which sensor/group detected each object. A crosscheck compares sensor-specific/group-specific existence probabilities with fused probabilities and sensor detection probabilities to identify false negative and false positive cases for each sensor or sensor group. Specifically, these identified false negative or false positive cases, assigned to individual objects, are stored in either a false negative list or a false positive list. This storage is done specifically for the sensor or sensor group that produced the error, and it considers the object's position within the fusion list.
3. The method as recited in claim 1, wherein in the crosscheck according to step e), the following are compared to one another: the sensor-specific or sensor-group-specific existence probabilities, additional information of the sensor or the sensor group, fused existence probabilities, and the sensor detection probabilities.
This method enhances the estimation of object existence probabilities by detecting objects using vehicle/infrastructure sensors, estimating their individual existence probabilities via detection models, and merging them into a fusion list where each object has a state and an existence probability. The system also tracks which sensor or sensor group detected each object in the latest cycle. A crucial part of this method is a crosscheck that identifies false negative and false positive cases for each sensor or sensor group. During this crosscheck, several factors are compared: the existence probabilities specific to each sensor or sensor group, any additional information provided by that sensor or sensor group, the overall fused existence probabilities from the merged list, and the sensor's individual detection probabilities.
4. The method as recited in claim 1, wherein after cycling through the method steps a) through g), a number of false negative cases and a number of false positive cases is reduced.
This method aims to improve the estimation of object existence probabilities. It begins by detecting objects with vehicle or infrastructure sensors, then each sensor/tracker estimates an object's status and its existence probability using a detection probability model. These detected objects are merged into a fusion list, each receiving a state and an existence probability, along with information on which sensor or sensor group detected them in the last cycle. A crosscheck then compares sensor-specific/group-specific existence probabilities with fused probabilities and sensor detection probabilities to identify false negative and false positive cases for each sensor or sensor group. A key result of repeatedly applying these method steps is a reduction in the overall number of false negative cases and false positive cases detected.
5. The method as recited in claim 1, wherein through consideration of the false negative cases and false positive cases fed back via the at least one feedback branch to the detection probability model and the clutter probability model, a first assumption of an existence probability of an object is transformed into a more accurate assumption of the existence probability.
This method improves the estimation of object existence probabilities by detecting objects with vehicle/infrastructure sensors, estimating their probabilities using detection models, and merging them into a fusion list, complete with detection source information. A crosscheck identifies false negative and false positive cases for each sensor or sensor group. Crucially, these identified false negative and false positive cases are fed back into the system via at least one feedback mechanism. This feedback specifically adjusts and refines both the detection probability model (used by individual sensors) and a clutter probability model. By continuously incorporating this feedback, an initial assumption of an object's existence probability is iteratively refined, leading to a significantly more accurate estimation of that existence probability.
6. The method as recited in claim 1, wherein an accuracy of an assumed detection probability of a first sensor is improved by sensor-specific false negative cases or false positive cases for the first sensor.
This method improves the estimation of object existence probabilities. It involves detecting objects with vehicle or infrastructure sensors, where each sensor/tracker estimates an object's status and its existence probability using a detection probability model. Detected objects are merged into a fusion list, assigned a state and existence probability, along with information on which sensor or sensor group detected them. A crosscheck then compares sensor-specific/group-specific existence probabilities with fused probabilities and sensor detection probabilities to identify false negative and false positive cases for each sensor or sensor group. A specific aspect of this method is that the accuracy of an assumed detection probability for a particular sensor is enhanced by incorporating the false negative cases or false positive cases that are unique to that first sensor.
7. The method as recited in claim 1, wherein an accuracy of an assumed detection probability of a first sensor group is improved by sensor-group-specific false negative cases or false positive cases for the first sensor group.
This method improves the estimation of object existence probabilities. Objects are detected using sensors installed in a vehicle and/or an infrastructure component, with each sensor or tracker estimating an object's status and existence probability via a detection probability model. Detected objects are merged into a fusion list, where each is assigned a state, an existence probability, and details on which sensor or sensor group detected it in the last cycle. A crosscheck compares sensor-specific/group-specific existence probabilities with fused probabilities and sensor detection probabilities to identify false negative and false positive cases for each sensor or sensor group. Importantly, the accuracy of an assumed detection probability for a particular sensor group is improved by using the false negative or false positive cases specifically identified for that sensor group.
8. The method as recited in claim 1, wherein the method improves an accuracy of the modeling of existence probabilities of objects detected by the sensors and/or the sensor groups.
This method targets improving the estimation of object existence probabilities. It functions by detecting objects using sensors in a vehicle and/or an infrastructure component. Each sensor or sensor group's tracker estimates an object's status and its existence probability using a detection probability model. These detected objects are then merged into a unified fusion list, with each object assigned a state, an existence probability, and additional information indicating its detection source in the last cycle. A crosscheck systematically compares sensor-specific/group-specific existence probabilities with fused probabilities and sensor detection probabilities to ascertain false negative and false positive cases for each sensor or sensor group. Ultimately, the entire method significantly improves the accuracy of how existence probabilities are modeled for all objects detected by the sensors and/or sensor groups.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
July 28, 2022
March 19, 2024
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.